X-Git-Url: http://drtracing.org/?a=blobdiff_plain;f=fs%2Fbtrfs%2Fordered-data.c;h=676e4bd65c529cb3332da5a62679a44ed5439bcd;hb=3eaa2885276fd6dac7b076a793932428b7168e74;hp=5e4c0d95ce43224bad1f9c94655986f571f7c089;hpb=594a24eb0e7fa8413f8b443863be4b7c72bfde9f;p=deliverable%2Flinux.git

diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c
index 5e4c0d95ce43..676e4bd65c52 100644
--- a/fs/btrfs/ordered-data.c
+++ b/fs/btrfs/ordered-data.c
@@ -19,51 +19,35 @@
 #include <linux/gfp.h>
 #include <linux/slab.h>
 #include <linux/blkdev.h>
+#include <linux/writeback.h>
+#include <linux/pagevec.h>
 #include "ctree.h"
 #include "transaction.h"
 #include "btrfs_inode.h"
+#include "extent_io.h"
 
-struct tree_entry {
-	u64 root_objectid;
-	u64 objectid;
-	struct inode *inode;
-	struct rb_node rb_node;
-};
 
-/*
- * returns > 0 if entry passed (root, objectid) is > entry,
- * < 0 if (root, objectid) < entry and zero if they are equal
- */
-static int comp_entry(struct tree_entry *entry, u64 root_objectid,
-		      u64 objectid)
+static u64 entry_end(struct btrfs_ordered_extent *entry)
 {
-	if (root_objectid < entry->root_objectid)
-		return -1;
-	if (root_objectid > entry->root_objectid)
-		return 1;
-	if (objectid < entry->objectid)
-		return -1;
-	if (objectid > entry->objectid)
-		return 1;
-	return 0;
+	if (entry->file_offset + entry->len < entry->file_offset)
+		return (u64)-1;
+	return entry->file_offset + entry->len;
 }
 
-static struct rb_node *tree_insert(struct rb_root *root, u64 root_objectid,
-				   u64 objectid, struct rb_node *node)
+static struct rb_node *tree_insert(struct rb_root *root, u64 file_offset,
+				   struct rb_node *node)
 {
 	struct rb_node ** p = &root->rb_node;
 	struct rb_node * parent = NULL;
-	struct tree_entry *entry;
-	int comp;
+	struct btrfs_ordered_extent *entry;
 
 	while(*p) {
 		parent = *p;
-		entry = rb_entry(parent, struct tree_entry, rb_node);
+		entry = rb_entry(parent, struct btrfs_ordered_extent, rb_node);
 
-		comp = comp_entry(entry, root_objectid, objectid);
-		if (comp < 0)
+		if (file_offset < entry->file_offset)
 			p = &(*p)->rb_left;
-		else if (comp > 0)
+		else if (file_offset >= entry_end(entry))
 			p = &(*p)->rb_right;
 		else
 			return parent;
@@ -74,24 +58,23 @@ static struct rb_node *tree_insert(struct rb_root *root, u64 root_objectid,
 	return NULL;
 }
 
-static struct rb_node *__tree_search(struct rb_root *root, u64 root_objectid,
-				     u64 objectid, struct rb_node **prev_ret)
+static struct rb_node *__tree_search(struct rb_root *root, u64 file_offset,
+				     struct rb_node **prev_ret)
 {
 	struct rb_node * n = root->rb_node;
 	struct rb_node *prev = NULL;
-	struct tree_entry *entry;
-	struct tree_entry *prev_entry = NULL;
-	int comp;
+	struct rb_node *test;
+	struct btrfs_ordered_extent *entry;
+	struct btrfs_ordered_extent *prev_entry = NULL;
 
 	while(n) {
-		entry = rb_entry(n, struct tree_entry, rb_node);
+		entry = rb_entry(n, struct btrfs_ordered_extent, rb_node);
 		prev = n;
 		prev_entry = entry;
-		comp = comp_entry(entry, root_objectid, objectid);
 
-		if (comp < 0)
+		if (file_offset < entry->file_offset)
 			n = n->rb_left;
-		else if (comp > 0)
+		else if (file_offset >= entry_end(entry))
 			n = n->rb_right;
 		else
 			return n;
@@ -99,194 +82,618 @@ static struct rb_node *__tree_search(struct rb_root *root, u64 root_objectid,
 	if (!prev_ret)
 		return NULL;
 
-	while(prev && comp_entry(prev_entry, root_objectid, objectid) >= 0) {
-		prev = rb_next(prev);
-		prev_entry = rb_entry(prev, struct tree_entry, rb_node);
+	while(prev && file_offset >= entry_end(prev_entry)) {
+		test = rb_next(prev);
+		if (!test)
+			break;
+		prev_entry = rb_entry(test, struct btrfs_ordered_extent,
+				      rb_node);
+		if (file_offset < entry_end(prev_entry))
+			break;
+
+		prev = test;
+	}
+	if (prev)
+		prev_entry = rb_entry(prev, struct btrfs_ordered_extent,
+				      rb_node);
+	while(prev && file_offset < entry_end(prev_entry)) {
+		test = rb_prev(prev);
+		if (!test)
+			break;
+		prev_entry = rb_entry(test, struct btrfs_ordered_extent,
+				      rb_node);
+		prev = test;
 	}
 	*prev_ret = prev;
 	return NULL;
 }
 
-static inline struct rb_node *tree_search(struct rb_root *root,
-					  u64 root_objectid, u64 objectid)
+static int offset_in_entry(struct btrfs_ordered_extent *entry, u64 file_offset)
 {
+	if (file_offset < entry->file_offset ||
+	    entry->file_offset + entry->len <= file_offset)
+		return 0;
+	return 1;
+}
+
+static inline struct rb_node *tree_search(struct btrfs_ordered_inode_tree *tree,
+					  u64 file_offset)
+{
+	struct rb_root *root = &tree->tree;
 	struct rb_node *prev;
 	struct rb_node *ret;
-	ret = __tree_search(root, root_objectid, objectid, &prev);
+	struct btrfs_ordered_extent *entry;
+
+	if (tree->last) {
+		entry = rb_entry(tree->last, struct btrfs_ordered_extent,
+				 rb_node);
+		if (offset_in_entry(entry, file_offset))
+			return tree->last;
+	}
+	ret = __tree_search(root, file_offset, &prev);
 	if (!ret)
-		return prev;
+		ret = prev;
+	if (ret)
+		tree->last = ret;
 	return ret;
 }
 
-int btrfs_add_ordered_inode(struct inode *inode)
+/* allocate and add a new ordered_extent into the per-inode tree.
+ * file_offset is the logical offset in the file
+ *
+ * start is the disk block number of an extent already reserved in the
+ * extent allocation tree
+ *
+ * len is the length of the extent
+ *
+ * This also sets the EXTENT_ORDERED bit on the range in the inode.
+ *
+ * The tree is given a single reference on the ordered extent that was
+ * inserted.
+ */
+int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
+			     u64 start, u64 len)
 {
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	u64 root_objectid = root->root_key.objectid;
-	u64 transid = root->fs_info->running_transaction->transid;
-	struct tree_entry *entry;
-	struct rb_node *node;
 	struct btrfs_ordered_inode_tree *tree;
+	struct rb_node *node;
+	struct btrfs_ordered_extent *entry;
 
-	if (transid <= BTRFS_I(inode)->ordered_trans)
-		return 0;
+	tree = &BTRFS_I(inode)->ordered_tree;
+	entry = kzalloc(sizeof(*entry), GFP_NOFS);
+	if (!entry)
+		return -ENOMEM;
+
+	mutex_lock(&tree->mutex);
+	entry->file_offset = file_offset;
+	entry->start = start;
+	entry->len = len;
+	entry->inode = inode;
 
-	tree = &root->fs_info->running_transaction->ordered_inode_tree;
+	/* one ref for the tree */
+	atomic_set(&entry->refs, 1);
+	init_waitqueue_head(&entry->wait);
+	INIT_LIST_HEAD(&entry->list);
+	INIT_LIST_HEAD(&entry->root_extent_list);
 
-	read_lock(&tree->lock);
-	node = __tree_search(&tree->tree, root_objectid, inode->i_ino, NULL);
-	read_unlock(&tree->lock);
+	node = tree_insert(&tree->tree, file_offset,
+			   &entry->rb_node);
 	if (node) {
-		return 0;
+		printk("warning dup entry from add_ordered_extent\n");
+		BUG();
 	}
+	set_extent_ordered(&BTRFS_I(inode)->io_tree, file_offset,
+			   entry_end(entry) - 1, GFP_NOFS);
 
-	entry = kmalloc(sizeof(*entry), GFP_NOFS);
-	if (!entry)
-		return -ENOMEM;
+	spin_lock(&BTRFS_I(inode)->root->fs_info->ordered_extent_lock);
+	list_add_tail(&entry->root_extent_list,
+		      &BTRFS_I(inode)->root->fs_info->ordered_extents);
+	spin_unlock(&BTRFS_I(inode)->root->fs_info->ordered_extent_lock);
 
-	write_lock(&tree->lock);
-	entry->objectid = inode->i_ino;
-	entry->root_objectid = root_objectid;
-	entry->inode = inode;
+	mutex_unlock(&tree->mutex);
+	BUG_ON(node);
+	return 0;
+}
 
-	node = tree_insert(&tree->tree, root_objectid,
-			   inode->i_ino, &entry->rb_node);
+/*
+ * Add a struct btrfs_ordered_sum into the list of checksums to be inserted
+ * when an ordered extent is finished.  If the list covers more than one
+ * ordered extent, it is split across multiples.
+ */
+int btrfs_add_ordered_sum(struct inode *inode,
+			  struct btrfs_ordered_extent *entry,
+			  struct btrfs_ordered_sum *sum)
+{
+	struct btrfs_ordered_inode_tree *tree;
+
+	tree = &BTRFS_I(inode)->ordered_tree;
+	mutex_lock(&tree->mutex);
+	list_add_tail(&sum->list, &entry->list);
+	mutex_unlock(&tree->mutex);
+	return 0;
+}
+
+/*
+ * this is used to account for finished IO across a given range
+ * of the file.  The IO should not span ordered extents.  If
+ * a given ordered_extent is completely done, 1 is returned, otherwise
+ * 0.
+ *
+ * test_and_set_bit on a flag in the struct btrfs_ordered_extent is used
+ * to make sure this function only returns 1 once for a given ordered extent.
+ */
+int btrfs_dec_test_ordered_pending(struct inode *inode,
+				   u64 file_offset, u64 io_size)
+{
+	struct btrfs_ordered_inode_tree *tree;
+	struct rb_node *node;
+	struct btrfs_ordered_extent *entry;
+	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+	int ret;
+
+	tree = &BTRFS_I(inode)->ordered_tree;
+	mutex_lock(&tree->mutex);
+	clear_extent_ordered(io_tree, file_offset, file_offset + io_size - 1,
+			     GFP_NOFS);
+	node = tree_search(tree, file_offset);
+	if (!node) {
+		ret = 1;
+		goto out;
+	}
+
+	entry = rb_entry(node, struct btrfs_ordered_extent, rb_node);
+	if (!offset_in_entry(entry, file_offset)) {
+		ret = 1;
+		goto out;
+	}
 
-	BTRFS_I(inode)->ordered_trans = transid;
+	ret = test_range_bit(io_tree, entry->file_offset,
+			     entry->file_offset + entry->len - 1,
+			     EXTENT_ORDERED, 0);
+	if (ret == 0)
+		ret = test_and_set_bit(BTRFS_ORDERED_IO_DONE, &entry->flags);
+out:
+	mutex_unlock(&tree->mutex);
+	return ret == 0;
+}
 
-	write_unlock(&tree->lock);
-	if (node)
+/*
+ * used to drop a reference on an ordered extent.  This will free
+ * the extent if the last reference is dropped
+ */
+int btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry)
+{
+	struct list_head *cur;
+	struct btrfs_ordered_sum *sum;
+
+	if (atomic_dec_and_test(&entry->refs)) {
+		while(!list_empty(&entry->list)) {
+			cur = entry->list.next;
+			sum = list_entry(cur, struct btrfs_ordered_sum, list);
+			list_del(&sum->list);
+			kfree(sum);
+		}
 		kfree(entry);
-	else
-		igrab(inode);
+	}
 	return 0;
 }
 
-int btrfs_find_first_ordered_inode(struct btrfs_ordered_inode_tree *tree,
-				   u64 *root_objectid, u64 *objectid,
-				   struct inode **inode)
+/*
+ * remove an ordered extent from the tree.  No references are dropped
+ * but, anyone waiting on this extent is woken up.
+ */
+int btrfs_remove_ordered_extent(struct inode *inode,
+				struct btrfs_ordered_extent *entry)
 {
-	struct tree_entry *entry;
+	struct btrfs_ordered_inode_tree *tree;
 	struct rb_node *node;
 
-	write_lock(&tree->lock);
-	node = tree_search(&tree->tree, *root_objectid, *objectid);
-	if (!node) {
-		write_unlock(&tree->lock);
-		return 0;
+	tree = &BTRFS_I(inode)->ordered_tree;
+	mutex_lock(&tree->mutex);
+	node = &entry->rb_node;
+	rb_erase(node, &tree->tree);
+	tree->last = NULL;
+	set_bit(BTRFS_ORDERED_COMPLETE, &entry->flags);
+
+	spin_lock(&BTRFS_I(inode)->root->fs_info->ordered_extent_lock);
+	list_del_init(&entry->root_extent_list);
+	spin_unlock(&BTRFS_I(inode)->root->fs_info->ordered_extent_lock);
+
+	mutex_unlock(&tree->mutex);
+	wake_up(&entry->wait);
+	return 0;
+}
+
+int btrfs_wait_ordered_extents(struct btrfs_root *root)
+{
+	struct list_head splice;
+	struct list_head *cur;
+	struct btrfs_ordered_extent *ordered;
+	struct inode *inode;
+
+	INIT_LIST_HEAD(&splice);
+
+	spin_lock(&root->fs_info->ordered_extent_lock);
+	list_splice_init(&root->fs_info->ordered_extents, &splice);
+	while(!list_empty(&splice)) {
+		cur = splice.next;
+		ordered = list_entry(cur, struct btrfs_ordered_extent,
+				     root_extent_list);
+		list_del_init(&ordered->root_extent_list);
+		atomic_inc(&ordered->refs);
+		inode = ordered->inode;
+
+		/*
+		 * the inode can't go away until all the pages are gone
+		 * and the pages won't go away while there is still
+		 * an ordered extent and the ordered extent won't go
+		 * away until it is off this list.  So, we can safely
+		 * increment i_count here and call iput later
+		 */
+		atomic_inc(&inode->i_count);
+		spin_unlock(&root->fs_info->ordered_extent_lock);
+
+		btrfs_start_ordered_extent(inode, ordered, 1);
+		btrfs_put_ordered_extent(ordered);
+		iput(inode);
+
+		spin_lock(&root->fs_info->ordered_extent_lock);
 	}
-	entry = rb_entry(node, struct tree_entry, rb_node);
+	spin_unlock(&root->fs_info->ordered_extent_lock);
+	return 0;
+}
 
-	while(comp_entry(entry, *root_objectid, *objectid) >= 0) {
-		node = rb_next(node);
-		if (!node)
+/*
+ * Used to start IO or wait for a given ordered extent to finish.
+ *
+ * If wait is one, this effectively waits on page writeback for all the pages
+ * in the extent, and it waits on the io completion code to insert
+ * metadata into the btree corresponding to the extent
+ */
+void btrfs_start_ordered_extent(struct inode *inode,
+				       struct btrfs_ordered_extent *entry,
+				       int wait)
+{
+	u64 start = entry->file_offset;
+	u64 end = start + entry->len - 1;
+
+	/*
+	 * pages in the range can be dirty, clean or writeback.  We
+	 * start IO on any dirty ones so the wait doesn't stall waiting
+	 * for pdflush to find them
+	 */
+	btrfs_fdatawrite_range(inode->i_mapping, start, end, WB_SYNC_NONE);
+	if (wait)
+		wait_event(entry->wait, test_bit(BTRFS_ORDERED_COMPLETE,
+						 &entry->flags));
+}
+
+/*
+ * Used to wait on ordered extents across a large range of bytes.
+ */
+void btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len)
+{
+	u64 end;
+	u64 orig_end;
+	u64 wait_end;
+	struct btrfs_ordered_extent *ordered;
+
+	if (start + len < start) {
+		orig_end = INT_LIMIT(loff_t);
+	} else {
+		orig_end = start + len - 1;
+		if (orig_end > INT_LIMIT(loff_t))
+			orig_end = INT_LIMIT(loff_t);
+	}
+	wait_end = orig_end;
+again:
+	/* start IO across the range first to instantiate any delalloc
+	 * extents
+	 */
+	btrfs_fdatawrite_range(inode->i_mapping, start, orig_end, WB_SYNC_NONE);
+
+	btrfs_wait_on_page_writeback_range(inode->i_mapping,
+					   start >> PAGE_CACHE_SHIFT,
+					   orig_end >> PAGE_CACHE_SHIFT);
+
+	end = orig_end;
+	while(1) {
+		ordered = btrfs_lookup_first_ordered_extent(inode, end);
+		if (!ordered) {
 			break;
-		entry = rb_entry(node, struct tree_entry, rb_node);
+		}
+		if (ordered->file_offset > orig_end) {
+			btrfs_put_ordered_extent(ordered);
+			break;
+		}
+		if (ordered->file_offset + ordered->len < start) {
+			btrfs_put_ordered_extent(ordered);
+			break;
+		}
+		btrfs_start_ordered_extent(inode, ordered, 1);
+		end = ordered->file_offset;
+		btrfs_put_ordered_extent(ordered);
+		if (end == 0 || end == start)
+			break;
+		end--;
 	}
-	if (!node) {
-		write_unlock(&tree->lock);
-		return 0;
+	if (test_range_bit(&BTRFS_I(inode)->io_tree, start, orig_end,
+			   EXTENT_ORDERED | EXTENT_DELALLOC, 0)) {
+		printk("inode %lu still ordered or delalloc after wait "
+		       "%llu %llu\n", inode->i_ino,
+		       (unsigned long long)start,
+		       (unsigned long long)orig_end);
+		goto again;
 	}
+}
 
-	*root_objectid = entry->root_objectid;
-	*inode = entry->inode;
-	atomic_inc(&entry->inode->i_count);
-	*objectid = entry->objectid;
-	write_unlock(&tree->lock);
-	return 1;
+/*
+ * find an ordered extent corresponding to file_offset.  return NULL if
+ * nothing is found, otherwise take a reference on the extent and return it
+ */
+struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct inode *inode,
+							 u64 file_offset)
+{
+	struct btrfs_ordered_inode_tree *tree;
+	struct rb_node *node;
+	struct btrfs_ordered_extent *entry = NULL;
+
+	tree = &BTRFS_I(inode)->ordered_tree;
+	mutex_lock(&tree->mutex);
+	node = tree_search(tree, file_offset);
+	if (!node)
+		goto out;
+
+	entry = rb_entry(node, struct btrfs_ordered_extent, rb_node);
+	if (!offset_in_entry(entry, file_offset))
+		entry = NULL;
+	if (entry)
+		atomic_inc(&entry->refs);
+out:
+	mutex_unlock(&tree->mutex);
+	return entry;
 }
 
-int btrfs_find_del_first_ordered_inode(struct btrfs_ordered_inode_tree *tree,
-				       u64 *root_objectid, u64 *objectid,
-				       struct inode **inode)
+/*
+ * lookup and return any extent before 'file_offset'.  NULL is returned
+ * if none is found
+ */
+struct btrfs_ordered_extent *
+btrfs_lookup_first_ordered_extent(struct inode * inode, u64 file_offset)
 {
-	struct tree_entry *entry;
+	struct btrfs_ordered_inode_tree *tree;
 	struct rb_node *node;
+	struct btrfs_ordered_extent *entry = NULL;
+
+	tree = &BTRFS_I(inode)->ordered_tree;
+	mutex_lock(&tree->mutex);
+	node = tree_search(tree, file_offset);
+	if (!node)
+		goto out;
+
+	entry = rb_entry(node, struct btrfs_ordered_extent, rb_node);
+	atomic_inc(&entry->refs);
+out:
+	mutex_unlock(&tree->mutex);
+	return entry;
+}
 
-	write_lock(&tree->lock);
-	node = tree_search(&tree->tree, *root_objectid, *objectid);
-	if (!node) {
-		write_unlock(&tree->lock);
-		return 0;
+/*
+ * After an extent is done, call this to conditionally update the on disk
+ * i_size.  i_size is updated to cover any fully written part of the file.
+ */
+int btrfs_ordered_update_i_size(struct inode *inode,
+				struct btrfs_ordered_extent *ordered)
+{
+	struct btrfs_ordered_inode_tree *tree = &BTRFS_I(inode)->ordered_tree;
+	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+	u64 disk_i_size;
+	u64 new_i_size;
+	u64 i_size_test;
+	struct rb_node *node;
+	struct btrfs_ordered_extent *test;
+
+	mutex_lock(&tree->mutex);
+	disk_i_size = BTRFS_I(inode)->disk_i_size;
+
+	/*
+	 * if the disk i_size is already at the inode->i_size, or
+	 * this ordered extent is inside the disk i_size, we're done
+	 */
+	if (disk_i_size >= inode->i_size ||
+	    ordered->file_offset + ordered->len <= disk_i_size) {
+		goto out;
 	}
 
-	entry = rb_entry(node, struct tree_entry, rb_node);
-	while(comp_entry(entry, *root_objectid, *objectid) >= 0) {
-		node = rb_next(node);
+	/*
+	 * we can't update the disk_isize if there are delalloc bytes
+	 * between disk_i_size and  this ordered extent
+	 */
+	if (test_range_bit(io_tree, disk_i_size,
+			   ordered->file_offset + ordered->len - 1,
+			   EXTENT_DELALLOC, 0)) {
+		goto out;
+	}
+	/*
+	 * walk backward from this ordered extent to disk_i_size.
+	 * if we find an ordered extent then we can't update disk i_size
+	 * yet
+	 */
+	node = &ordered->rb_node;
+	while(1) {
+		node = rb_prev(node);
 		if (!node)
 			break;
-		entry = rb_entry(node, struct tree_entry, rb_node);
+		test = rb_entry(node, struct btrfs_ordered_extent, rb_node);
+		if (test->file_offset + test->len <= disk_i_size)
+			break;
+		if (test->file_offset >= inode->i_size)
+			break;
+		if (test->file_offset >= disk_i_size)
+			goto out;
 	}
-	if (!node) {
-		write_unlock(&tree->lock);
-		return 0;
+	new_i_size = min_t(u64, entry_end(ordered), i_size_read(inode));
+
+	/*
+	 * at this point, we know we can safely update i_size to at least
+	 * the offset from this ordered extent.  But, we need to
+	 * walk forward and see if ios from higher up in the file have
+	 * finished.
+	 */
+	node = rb_next(&ordered->rb_node);
+	i_size_test = 0;
+	if (node) {
+		/*
+		 * do we have an area where IO might have finished
+		 * between our ordered extent and the next one.
+		 */
+		test = rb_entry(node, struct btrfs_ordered_extent, rb_node);
+		if (test->file_offset > entry_end(ordered)) {
+			i_size_test = test->file_offset - 1;
+		}
+	} else {
+		i_size_test = i_size_read(inode);
 	}
 
-	*root_objectid = entry->root_objectid;
-	*objectid = entry->objectid;
-	*inode = entry->inode;
-	atomic_inc(&entry->inode->i_count);
-	rb_erase(node, &tree->tree);
-	write_unlock(&tree->lock);
-	kfree(entry);
-	return 1;
+	/*
+	 * i_size_test is the end of a region after this ordered
+	 * extent where there are no ordered extents.  As long as there
+	 * are no delalloc bytes in this area, it is safe to update
+	 * disk_i_size to the end of the region.
+	 */
+	if (i_size_test > entry_end(ordered) &&
+	    !test_range_bit(io_tree, entry_end(ordered), i_size_test,
+			   EXTENT_DELALLOC, 0)) {
+		new_i_size = min_t(u64, i_size_test, i_size_read(inode));
+	}
+	BTRFS_I(inode)->disk_i_size = new_i_size;
+out:
+	mutex_unlock(&tree->mutex);
+	return 0;
 }
 
-static void __btrfs_del_ordered_inode(struct btrfs_ordered_inode_tree *tree,
-				     struct inode *inode,
-				     u64 root_objectid, u64 objectid)
+/*
+ * search the ordered extents for one corresponding to 'offset' and
+ * try to find a checksum.  This is used because we allow pages to
+ * be reclaimed before their checksum is actually put into the btree
+ */
+int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u32 *sum)
 {
-	struct tree_entry *entry;
-	struct rb_node *node;
-	struct rb_node *prev;
+	struct btrfs_ordered_sum *ordered_sum;
+	struct btrfs_sector_sum *sector_sums;
+	struct btrfs_ordered_extent *ordered;
+	struct btrfs_ordered_inode_tree *tree = &BTRFS_I(inode)->ordered_tree;
+	struct list_head *cur;
+	unsigned long num_sectors;
+	unsigned long i;
+	u32 sectorsize = BTRFS_I(inode)->root->sectorsize;
+	int ret = 1;
+
+	ordered = btrfs_lookup_ordered_extent(inode, offset);
+	if (!ordered)
+		return 1;
 
-	write_lock(&tree->lock);
-	node = __tree_search(&tree->tree, root_objectid, objectid, &prev);
-	if (!node) {
-		write_unlock(&tree->lock);
-		return;
+	mutex_lock(&tree->mutex);
+	list_for_each_prev(cur, &ordered->list) {
+		ordered_sum = list_entry(cur, struct btrfs_ordered_sum, list);
+		if (offset >= ordered_sum->file_offset) {
+			num_sectors = ordered_sum->len / sectorsize;
+			sector_sums = ordered_sum->sums;
+			for (i = 0; i < num_sectors; i++) {
+				if (sector_sums[i].offset == offset) {
+					*sum = sector_sums[i].sum;
+					ret = 0;
+					goto out;
+				}
+			}
+		}
 	}
-	rb_erase(node, &tree->tree);
-	BTRFS_I(inode)->ordered_trans = 0;
-	write_unlock(&tree->lock);
-	atomic_dec(&inode->i_count);
-	entry = rb_entry(node, struct tree_entry, rb_node);
-	kfree(entry);
-	return;
+out:
+	mutex_unlock(&tree->mutex);
+	btrfs_put_ordered_extent(ordered);
+	return ret;
 }
 
-void btrfs_del_ordered_inode(struct inode *inode, int force)
+
+/**
+ * taken from mm/filemap.c because it isn't exported
+ *
+ * __filemap_fdatawrite_range - start writeback on mapping dirty pages in range
+ * @mapping:	address space structure to write
+ * @start:	offset in bytes where the range starts
+ * @end:	offset in bytes where the range ends (inclusive)
+ * @sync_mode:	enable synchronous operation
+ *
+ * Start writeback against all of a mapping's dirty pages that lie
+ * within the byte offsets <start, end> inclusive.
+ *
+ * If sync_mode is WB_SYNC_ALL then this is a "data integrity" operation, as
+ * opposed to a regular memory cleansing writeback.  The difference between
+ * these two operations is that if a dirty page/buffer is encountered, it must
+ * be waited upon, and not just skipped over.
+ */
+int btrfs_fdatawrite_range(struct address_space *mapping, loff_t start,
+			   loff_t end, int sync_mode)
 {
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	u64 root_objectid = root->root_key.objectid;
+	struct writeback_control wbc = {
+		.sync_mode = sync_mode,
+		.nr_to_write = mapping->nrpages * 2,
+		.range_start = start,
+		.range_end = end,
+		.for_writepages = 1,
+	};
+	return btrfs_writepages(mapping, &wbc);
+}
 
-	if (!BTRFS_I(inode)->ordered_trans) {
-		return;
-	}
+/**
+ * taken from mm/filemap.c because it isn't exported
+ *
+ * wait_on_page_writeback_range - wait for writeback to complete
+ * @mapping:	target address_space
+ * @start:	beginning page index
+ * @end:	ending page index
+ *
+ * Wait for writeback to complete against pages indexed by start->end
+ * inclusive
+ */
+int btrfs_wait_on_page_writeback_range(struct address_space *mapping,
+				       pgoff_t start, pgoff_t end)
+{
+	struct pagevec pvec;
+	int nr_pages;
+	int ret = 0;
+	pgoff_t index;
 
-	if (!force && (mapping_tagged(inode->i_mapping, PAGECACHE_TAG_DIRTY) ||
-	    mapping_tagged(inode->i_mapping, PAGECACHE_TAG_WRITEBACK)))
-		return;
+	if (end < start)
+		return 0;
 
-	spin_lock(&root->fs_info->new_trans_lock);
-	if (root->fs_info->running_transaction) {
-		struct btrfs_ordered_inode_tree *tree;
-		tree = &root->fs_info->running_transaction->ordered_inode_tree;
-		 __btrfs_del_ordered_inode(tree, inode, root_objectid,
-						inode->i_ino);
+	pagevec_init(&pvec, 0);
+	index = start;
+	while ((index <= end) &&
+			(nr_pages = pagevec_lookup_tag(&pvec, mapping, &index,
+			PAGECACHE_TAG_WRITEBACK,
+			min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1)) != 0) {
+		unsigned i;
+
+		for (i = 0; i < nr_pages; i++) {
+			struct page *page = pvec.pages[i];
+
+			/* until radix tree lookup accepts end_index */
+			if (page->index > end)
+				continue;
+
+			wait_on_page_writeback(page);
+			if (PageError(page))
+				ret = -EIO;
+		}
+		pagevec_release(&pvec);
+		cond_resched();
 	}
-	spin_unlock(&root->fs_info->new_trans_lock);
-}
 
-int btrfs_ordered_throttle(struct btrfs_root *root, struct inode *inode)
-{
-	struct btrfs_transaction *cur = root->fs_info->running_transaction;
-	while(cur == root->fs_info->running_transaction &&
-	      atomic_read(&BTRFS_I(inode)->ordered_writeback)) {
-#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,18)
-		congestion_wait(WRITE, HZ/20);
-#else
-		blk_congestion_wait(WRITE, HZ/20);
-#endif
-	}
-	return 0;
+	/* Check for outstanding write errors */
+	if (test_and_clear_bit(AS_ENOSPC, &mapping->flags))
+		ret = -ENOSPC;
+	if (test_and_clear_bit(AS_EIO, &mapping->flags))
+		ret = -EIO;
+
+	return ret;
 }